Abstract Type 2 diabetes (T2D) is a highly prevalent disease for which new therapies are needed. The adipocyte differentiation and lipid storage pathways are involved in rare and common forms of diabetes and are targeted by thiazolidinediones (TZDs), which are efficacious but cause undesirable complications. Designing better therapies to target adipocyte differentiation/lipid storage is impeded by incomplete knowledge of which genes in these pathways are relevant to T2D in humans, or how they might be modulated to achieve therapeutic efficacy. Disease-associated rare coding variants directly identify human disease-relevant gene modulations, and our recent study of 45,231 exomes suggested that such associations are likely observable within many genes within the adipocyte differentiation and lipid storage pathways. However, rare variant associations require large datasets to detect, and methods are currently lacking to identify which observed associations are most likely to (a) represent causal links to disease and (b) act through effects on a pathway of interest. The proposed project will address these gaps under the hypothesis that T2D-susceptibility rare coding variants that modulate the adipocyte differentiation/lipid storage pathways should impair these processes in vitro and predispose in vivo to an ?insulin resistance signature? of higher T2D risk, higher BMI-adjusted fasting insulin levels, higher triglyceride levels, lower hip circumference, and lower HDL levels. Specific aim 1 hypothesizes that larger exome datasets will identify new gene-level rare coding variant T2D associations, and that prior knowledge of gene function should affect the likelihood each association is causal. Coding variants in 150K-600K exomes will be tested for association with T2D and insulin resistance, and each gene's probability of causal association will be calculated by a new method to account for its empirically estimated prior likelihood of association. Specific aim 2 hypothesizes that genes associated with an insulin resistance signature in vivo should have a higher likelihood of impairing adipocyte differentiation/lipid storage when ablated in vitro, and that within these genes, only variants that fail to complement effects observed in vitro should increase T2D risk in vivo. Fifty genes with gene-level T2D associations will be screened via loss-of-function experiments in human pre-adipocytes, and genetic complementation experiments will be conducted for 50 variants in each of 5 genes whose ablation impairs adipocyte differentiation/lipid storage. Significance: T2D-associated coding variants with in vitro effects on adipocyte differentiation or lipid storage would suggest molecular gene perturbations to protect from or treat T2D. These and all other results of the project will be made publicly accessible through the NIDDK-funded AMP-T2D Knowledge Portal. The proposed approaches also apply to other biological processes and diseases.